Am. J. Hum. Genet.

69:315–326, 2001

Functional Complementation of a Genetic Deficiency with Human

Artificial Chromosomes
José E. Mejı́a,1 Adrian Willmott,1 Elaine Levy,2 William C. Earnshaw,3 and Zoia Larin1
1
Institute of Molecular Medicine and 2Wellcome Trust Centre for Human Genetics, University of Oxford, Oxford; and 3Institute of Cell and
Molecular Biology, University of Edinburgh, Edinburgh

We have shown functional complementation of a genetic deficiency in human cultured cells, using artificial chro-
mosomes derived from cloned human genomic fragments. A 404-kb human-artificial-chromosome (HAC) vector,
consisting of 220 kb of alphoid DNA from the centromere of chromosome 17, human telomeres, and the hypo-
xanthine guanine phosphoribosyltransferase (HPRT) genomic locus, was transferred to HPRT-deficient HT1080
fibrosarcoma cells. We generated several cell lines with low-copy-number, megabase-sized HACs containing a
functional centromere and one or possibly several copies of the HPRT1 gene complementing the metabolic deficiency.
The HACs consisted of alternating alphoid and nonalphoid DNA segments derived only from the input DNA
(within the sensitivity limits of FISH detection), and the largest continuous alphoid segment was 158–250 kb. The
study of both the structure and mitotic stability of these HACs offers insights into the mechanisms of centromere
formation in synthetic chromosomes and will further the development of this human-gene-transfer technology.

Introduction gene-transfer vectors is now a possibility. Owing to their

Several groups have had success in generating stable hu-
man artificial chromosomes (HACs) in mammalian cells
either by telomere-directed fragmentation of endogenous
human chromosomes or from the introduction of cloned
human centromeric and telomeric sequences (Harring-
ton et al. 1997; Ikeno et al. 1998; Henning et al. 1999;
Mills et al. 1999; Ebersole et al. 2000; Yang et al. 2000).
Both strategies have determined (1) that a-satellite, or
alphoid, DNA, the major sequence component at human
centromeres, can generate a functional de novo centro-
mere in human cells and (2) that a minimal size is im-
portant for minichromosome stability (Yang et al. 2000).
In addition, transfection of defined constructs into hu-
man cells has shown that in the absence of telomeric
repeats a large array of alphoid DNA alone is sufficient
to generate stable circular artificial minichromosomes
(Ebersole et al. 2000). In human cells, characterization
of neocentromeres devoid of normal centromeric se-
quences has led to the question of whether sequence
alone is sufficient to determine a centromere; however,
de novo centromere activation resulting from an epi-
genetic mechanism has yet to be identified (Karpen and
Allshire 1997; Willard 1998; Choo 2000).
The development, via either strategy, of HACs as
Received March 26, 2001; accepted for publication June 11, 2001;
electronically published July 10, 2001.
Address for correspondence and reprints: Dr. Zoia Larin, Institute
of Molecular Medicine, John Radcliffe Hospital, Oxford OX3 9DS,
United Kingdom. E-mail: zlarin@molbiol.ox.ac.uk
䉷 2001 by The American Society of Human Genetics. All rights reserved.
0002-9297/2001/6902-0008$02.00
potential to carry entire genomic loci, HACs should
mediate long-term, tissue-specific expression of trans-
genes. Stable extrachromosomal maintenance of HACs
would overcome the problems of gene silencing result-
ing from random integration into the host genome and
should produce a regulated level of gene expression
throughout the lifetime of the cell. As a nonviral ap-
proach based on human functional elements, the use of
HACs may offer a safer alternative to viral therapeutic
gene-transfer methods currently being investigated, par-
ticularly in light of a major setback in a recent clinical
gene-therapy trial with adenoviral vectors (Willard
2000). Because of the size of HACs, however, methods
for their efficient delivery to human cells will have to
be developed.
A recently described approach to integrate one or a
number of genes into HACs involves the manipulation
of minichromosomes in chicken DT40 cells and the in-
troduction of defined regions containing human genes
by Cre-mediated recombination at loxP sites (Kuroiwa
et al. 2000). An alternative strategy is to incorporate,
into defined HAC vectors, either P1 artificial chromo-
some (PACs) or bacterial artificial chromosomes (BACs)
with large genomic fragments spanning an entire gene
locus and distant regulatory sequences and to introduce
the constructs into human cells. We constructed a HAC
vector for the transfer of a human marker gene, con-
taining alphoid DNA from the centromere of chro-
mosome 17, telomeric sequences, and, from human
Xq26.2, a large genomic segment spanning the hypo-
xanthine guanine phosphoribosyltransferase (HPRT)

315
316
locus, HPRT1 (MIM 308000; Project Ensembl) (Mejı́a
and Larin 2000). HPRT1 encodes a purine-salvage en-
zyme, and mutations in this gene result in Lesch-Nyhan
syndrome (MIM 300322), a neurodevelopmental dis-
order characterized by self-injurious and abnormal mo-
tor behavior (Caskey and Kruh 1979). In this study, we
introduced the HAC vector into HPRT-deficient human
fibrosarcoma cells and obtained stable artificial mini-
chromosomes that complemented the HPRT⫺ pheno-
type of the host cells. Our structural and functional
characterization of the artificial minichromosomes of-
fers insights into the requirements for centromere func-
tion and the prospects of using HACs as vectors for
gene transfer.
Material and Methods
HAC Construct, Enzymes, and Media
The assembly of the HAC construct used in this study,
JM2860, has been described in detail elsewhere (Mejı́a
and Larin 2000). b-Agarase and endonucleases—except
I-SceI, which was from Roche—were from New England
Biolabs. Cell-culture media, selective medium supple-
ments, and transfection reagents were purchased from
Gibco BRL unless otherwise indicated.
Lipofection of Cultured Cells
JM2860 DNA was prepared in plugs of low-melting-
point agarose (SeaPlaque; FMC), as described elsewhere
(Larin 1995). The DNA (1–3 mg/agarose plug) was lin-
earized by digestion by I-SceI endonuclease, washed in 10
mM Tris-HCl, 1 mM EDTA buffer at pH 8, and then
equilibrated in the same buffer, containing 50 mM low-
molecular-weight (∼2,000) polyethylenimine (PEI) (Al-
drich) (Boussif et al. 1995; Marschall et al. 1999). The
plugs were melted at 65⬚C, treated with b-agarase, and
diluted, to a DNA concentration of 0.5 mg/ml, with
OptiMEM. Lipofectin—7.5 mg in 2 ml of OptiMEM—
was added to 2 ml of the DNA dilution, and 8 ml of
this transfection medium was added to subconfluent cells
of an HPRT⫺ HT1080 subline (Rasheed et al. 1974;
Lugo and Baker 1983), in a 10-cm petri dish. The next
day, transfection medium was replaced with normal
growth medium composed of Dulbecco’s modified Ea-
gle’s medium (DMEM) supplemented with 10% fetal
bovine serum (Globepharm). Confluent cells were split
into six dishes, and selection was applied, after the cells
had again reached confluence, for 9 d. Selective medium
supplements were used as follows: puromycin (Sigma),
0.22–0.25 mg/ml; G418, 125–160 mg/ml; and hypoxan-
thine, aminopterin, and thymidine (HAT) supplement
composed of 0.1 mM hypoxanthine, 0.4 mM aminop-
terin, and 16 mM thymidine. Colonies were picked into
Am. J. Hum. Genet. 69:315–326, 2001
24-well plates, expanded, and screened by PCR and
FISH.
PCR Screening of Positive Clones
Primer pair GGCCTCTTCGCTATTACGC (JNC9,
vector) and CGGGAGAATCTTCACAGGAA (JNC10,
insert) and primer pair CGGGCCTCTTCGCTATTAC
(JNC11, vector) and ACGGGAGAATCTTCACAGGA
(JNC12, insert), which generated PCR products of 293
bp and 296 bp, respectively, were used to detect the
junction region between the alphoid insert and a residual
fragment of the pBeloBAC11 vector (Kim et al. 1996)
in JM2860 (fig. 1A).
6-Thioguanine (6-TG)–Sensitivity Assay
For each minichromosome or control cell line, 105
cells were seeded in a well of a 6-well plate and were
grown in DMEM supplemented with 10% dialyzed fetal
bovine serum and 5 mg 6-TG/ml (Sigma). Sensitivity to
6-TG was assessed by comparison both with identical
cultures in which 6-TG had been omitted and between
the minichromosome cell lines and the HPRT⫹ (6-TG
sensitive), as well as the parental HPRT⫺ (6-TG resis-
tant), HT1080 cell lines.
FISH of Metaphase Chromosomes
Metaphase chromosome spreads were prepared on
microscope slides and hybridized using modifications of
standard FISH protocols. Artificial chromosomes in
JM2860-positive clones were detected with a biotin-la-
beled chromosome 17 alphoid (17 a) probe (D17Z1;
Oncor) and a fluorescein-streptavidin conjugate (Vec-
tor). Selected clones were analyzed by dual hybridization
with the 17 a probe and the HPRT1-locus PAC from
which JM2860 was derived; the latter was labeled with
digoxigenin (DIG-Nick Translation Mix; Roche Diag-
nostics), and hybridization was performed in the pres-
ence of unlabeled human C 0 t1-fraction DNA. Further
dual-hybridization analyses were performed with the bi-
otin-labeled 17 a probe and a digoxigenin-labeled uni-
versal-human-telomere probe (Oncor). In dual hybridi-
zations, the biotin-labeled probe was detected with a
streptavidin-Cy3 conjugate (Sigma). In some applica-
tions, sensitivity was enhanced by the use of an addi-
tional layer (i.e., biotinylated anti-streptavidin antibody
from goat [Vector]), prior to a secondary layer of fluo-
rescent conjugate. Digoxigenin-labeled probes were de-
tected with a mouse monoclonal anti-digoxigenin anti-
body followed by fluorescein isothiocyanate (FITC)–
labeled sheep antibody anti–mouse immunoglobulins
(Ig) (Roche). When it was necessary, a final layer of
donkey anti–sheep IgG antibody and FITC conjugate
(Sigma) was used for increased sensitivity. The stringency
of hybridization conditions and washes was optimized
Mejı́a et al.: HAC-Mediated Rescue of HPRT Deficiency 317

Figure 1 Maps of the HAC construct. A, Schematic of JM2860 HAC construct (Mejı́a and Larin 2000) linearized with I-SceI endonuclease.
The construct contains both a 220-kb alphoid-DNA array from the human chromosome 17 centromere and a chromosome X segment with
the HPRT1 gene. The ends are capped by human-telomere repeats, and the vector arms contain selectable markers for mammalian-cell transfection
(PurR and neo), the ampicillin-resistance gene (AmpR) for selection in Escherichia coli, and lacZ as a reporter gene in mammalian cells. The
location of the region amplified by PCR in HAC DNA derived from positive clones is indicated below the map, together with cleavage sites
for the enzymes used in the structural analysis of the artificial minichromosomes (fig. 4A). B, Telomeric region of JM2860 prior to I-SceI
cleavage. The I-SceI sites downstream from each 1.2-kb telomeric repeat (Tel), the position of the XhoI restriction sites, and the repE and AmpR
gene probes used in the Bal31 analysis (fig. 5) are indicated below the map. The alphoid and chromosome X genomic inserts are denoted by
“17 a” and “Xq26.2,” respectively.

for every probe or probe combination, by varying the
concentration of formamide in the buffers. For micros-
copy, slides were mounted by use of Vectashield mount-
ing medium (Vector) containing, as fluorescent DNA
counterstains, propidium iodide or TO-PRO-3 iodide
(Molecular Probes). Fluorescence microscopy was per-
formed with an MRC-1024 laser-scanning confocal mi-
croscope and with LaserSharp software from Bio-Rad.
Chromosome Painting
A panel of 24 chromosome-specific, FITC-labeled
chromosome paint probes (Chromoprobe Multiprobe
Hybridisation System; Cytocell) were used in FISH, ac-
cording to the supplier’s protocol, and each paint was
combined with the aforementioned biotin-labeled 17 a
probe. Biotin detection and microscopic analysis of
slides were as described in the subsection “FISH of Meta-
phase Chromosomes.”
Fiber FISH
HT1080 cells were dissociated with trypsin and washed
in phosphate-buffered saline (PBS). Stretched DNA fibers
were prepared by first drying 10 ml of the cell suspension
in PBS, at 106 cells/ml, near the edge of a microscope slide
at 45–50⬚C. The slide was then placed vertically in a
slide–cover-plate assembly, and the cells were lysed down
the slide with 150 ml of 0.05 M NaOH in 28.6% ethanol.
This resulted in extended chromatin fibers, which were
affixed onto the slide with 200 ml of methanol, dehydrated
by successive washes in increasingly concentrated ethanol,
and, finally, air-dried. DNA fibers were denatured for 3
min in 70% formamide, 2# standard sodium citrate buf-
fer, at 70⬚C. FISH was performed with the 17 a and
HPRT1-locus probes as described in the subsection “FISH
of Metaphase Chromosomes,” and slides were mounted
for microscopy with 4’,6-diamidino-2-phenylindole as the
DNA counterstain.
Centromeric Protein C (CENP-C) Detection
Immunochemical detection of CENP-C was per-
formed as described elsewhere (Sullivan and Schwartz
1995), with slight modifications. In brief, colcemid-
treated cells were centrifuged onto microscope slides in
a Shandon Cytospin 2 apparatus and incubated with
rabbit anti–CENP-C antibodies (Saitoh et al. 1992) fol-
lowed by FITC-conjugated goat anti–rabbit Ig antibod-
ies. The slides were then formalin fixed, rinsed, and,
finally, treated with methanol:acetic acid (3:1) and air-
dried. After a formamide denaturation step of 9 min,
chromosome spreads were hybridized with the 17 a
probe, as described in the subsection “FISH of Meta-
phase Chromosomes.” The probe was detected with
streptavidin-Cy3 conjugate, with TO-PRO-3 iodide as
a DNA counterstain.
318
Restriction-Enzyme Digestion and Hybridization
Analysis
Genomic DNA from cultured cells was prepared in
low-melting-point agarose plugs (Larin 1995). Restric-
tion-enzyme digestions were performed overnight after
equilibration of the plugs in the buffer recommended by
the enzyme supplier, with 2 mM spermidine. The DNA
was fractionated by pulsed-field gel electrophoresis
(PFGE) in a CHEF Mapper apparatus (Bio-Rad). South-
ern blot transfer and hybridization analysis were per-
formed with [32P]-labeled DNA probes prepared with
the alphoid-DNA BAC from which the relevant con-
struct was derived.
g Irradiation of DNA
Genomic DNA purified in agarose plugs was exposed
to a 137Cs source (Gammacell 1000 irradiator; Atomic
Energy of Canada) for various times, to achieve the de-
sired doses of g irradiation (Taylor et al. 1996), and was
fractionated by PFGE, for Southern hybridization anal-
ysis. The probes used were pCMVb (Clontech), to detect
the lacZ and ampicillin-resistance genes in the JM2860
construct, and the 17 a–DNA BAC.
Bal31 Analysis
Genomic DNA purified in agarose plugs was digested
by Bal31 nuclease, as described elsewhere (Taylor et al.
1994), with BAL31 from New England Biolabs, ac-
cording to the manufacturer’s instructions. The DNA
was then restriction digested by XhoI and fractionated,
for Southern hybridization, by agarose-gel electropho-
resis. The blot was probed with the ampicillin-resistance
gene (1.5 kb) and a 0.35-kb PCR fragment from the
repE gene in the origin of replication of the BAC vector
(Mejı́a and Monaco 1997).
Results
HAC Construct
The circular 404-kb HAC construct, JM2860, was
generated as described in detail elsewhere (Mejı́a and
Larin 2000). Figure 1A shows a map of JM2860 after
specific linearization of the BAC with endonuclease I-
SceI. The construct consists of 220 kb of human 17 a
DNA, human telomeres at the end of each vector arm,
and, from human chromosome Xq26.2, a large genomic
fragment (162 kb) containing the entire HPRT1 genomic
locus. Reporter and selectable markers (lacZ as well as
neomycin- and puromycin-resistance genes) are present
in each vector arm, as shown (fig. 1).
Am. J. Hum. Genet. 69:315–326, 2001
Transfection of HPRT⫺ HT1080 Cells and Preliminary
Screening of Positive Clones
DNA from JM2860 was digested by I-SceI endonu-
clease and then, prior to transfer into HPRT⫺ HT1080
human fibrosarcoma cells by lipofection, condensed with
50 mM PEI (Boussif et al. 1995; Marschall et al. 1999).
Positive clones were selected in the presence of either (1)
G418 and puromycin, (2) HAT medium, or (3) G418,
puromycin, and HAT (i.e., a triple selection). DNA from
22 clones was initially screened via PCR using primers
spanning the junction region between the alphoid insert
and a residual fragment of the pBeloBAC11 vector (Kim
et al. 1996), 60 kb upstream from the HPRT1 gene in
JM2860 (fig. 1A). Seventeen clones showed the expected
PCR fragment (data not shown), suggesting that
JM2860 DNA had been transferred intact into HPRT⫺
HT1080 cells. To detect the presence of extrachromo-
somal elements, 14 independent clones were selected for
further investigation by FISH.
Analysis of Positive Clones by FISH and by
Immunofluorescence
Metaphase chromosomes prepared from 14 clones
were analyzed by FISH with a 17 a–DNA probe. Eleven
clones contained artificial minichromosomes present in
20%–100% of cells, with one or two copies per positive
cell. This represents a de novo minichromosome-forma-
tion frequency of ∼ 1 # 10⫺6, relative to the number of
HT1080 cells plated for transfection. Four cell lines—
AG1-1, AG3-1, AG6-1, and AP1-1—contained minichro-
mosomes only, and the remaining seven cell lines con-
tained minichromosomes and DNA that had also either
integrated into or truncated a host chromosome.
Further analysis of AG1-1, AG3-1, AG6-1, and AP1-
1, with both the 17 a probe and an HPRT1 probe,
showed that the two sequences colocalized on the arti-
ficial minichromosomes, and separate endogenous sig-
nals were detected, as expected, at the chromosome 17
centromeres and at the q arm of chromosome X (fig.
2A). Indirect immunofluorescence detection on meta-
phase chromosomes of an essential kinetochore protein,
CENP-C (Saitoh et al. 1992), followed by FISH with the
17 a probe, showed colocalization of the two signals on
the artificial minichromosomes and at the centromeres
of the endogenous chromosomes 17 (fig. 2B). As a
marker of functional centromeres, the CENP-C signal
was also detected at the centromeres of all endogenous
chromosomes as a doublet corresponding to the pair of
sister chromatids. In a proportion of minichromosomes,
CENP-C appeared as a singlet, probably because of their
small size relative to the endogenous chromosomes and
to the resolution of the microscopic analysis. The 17 a
probe and a human-telomere probe were also shown to
Mejı́a et al.: HAC-Mediated Rescue of HPRT Deficiency 319

Figure 2 FISH analysis of minichromosome cell lines. A–C, Details of metaphase chromosome spreads from four minichromosome cell
lines—AG1-1, AG3-1, AG6-1, and AP1-1—showing 17 a and DNA probe for the HPRT1 locus (A); 17 a and antibody to CENP-C, a marker
of functional centromeres (B); and 17 a– and human-telomeric–DNA probe. Note the colocalization, on artificial minichromosomes (arrowheads),
of the 17 a–DNA probe (red signal) and a second DNA or antibody probe (green signal). Overlap between the red and the green signals is
shown as white or yellow pseudocolor. Total DNA stained with TO-PRO-3 is shown as blue pseudocolor. In addition to doing so on minichromo-
somes, the probes gave expected signals on the endogenous chromosome 17 centromere (A–C), the HPRT1 locus on Xq26.2 (A), the kinetochores
of sister chromatids in each chromosome (B), and the telomeric termini of chromosomes (C). D, Dual hybridization of 17 a (red signal) and
HPRT1-locus (green signal) probes to stretched DNA fibers from the AG6-1 cell line. The signals show that both sequences alternate on the
minichromosome, as consecutive segments of irregular size. Gaps along the minichromosome fiber are expected where sequences are not
represented in the probes used—that is, in the 22-kb region of JM2860 represented in figure 1B and in repetitive sequences masked by human
C0t1 DNA used in the hybridization medium.

colocalize on the artificial minichromosomes (fig. 2C).
A striking visualization of the structure of the minichro-
mosomes in AG6-1 was provided by FISH analysis of
stretched DNA fibers prepared from metaphase cells.
Simultaneous hybridization with the 17 a and HPRT1
probes revealed alphoid-DNA and HPRT1 segments of
varying sizes alternating over the observed length of the
DNA fiber (fig. 2D).
Metaphase chromosomes from each cell line were also
hybridized, with a panel of 24 human-chromosome–
specific paint probes (chromosomes 1–22, X, and Y) in
conjunction with the 17 a probe, to determine whether
the HACs contained any endogenous chromosomal
DNA (fig. 3). Each paint probe detected the correspond-
ing endogenous chromosome, but only the 17 a probe
was observed to hybridize to the HACs and the endog-
enous chromosomes 17. We were unable to detect on
the HACs the presence of a signal from the X-chro-
mosome paint to the HPRT1 locus, presumably because
the signal is not within the detection limits of the paint
320 Am. J. Hum. Genet. 69:315–326, 2001

Figure 3 Analysis of minichromosomes by chromosome paints. Artificial minichromosomes (arrowheads), detected by the 17 a FISH
probe (red signal), are negative for a set of human-chromosome–specific (chromosomes 1–22, X, and Y) paint probes (green signal). In these
experiments, adequate detection of the corresponding natural chromosome is shown for each individual paint probe. The image shows details
of metaphase chromosome spreads from the AG6-1 cell line, representative of the results from three other minichromosome cell lines—AG1-
1, AG3-1, and AP1-1—analyzed in parallel.

probe. Overall, on the basis of both the paint-probe
results and the alternate structure revealed by fiber FISH,
we assume that the HACs did not contain any long
stretches of endogenous sequences within the sensitivity
level of each probe. In the course of the FISH analyses,
a truncation of the long arm of the Y chromosome was
observed in a portion of the cells. Rarely, other rear-
rangements were also observed, such as the translocation
of the q arm of chromosome 11, which is visible in figure
3; these rearrangements were not accompanied by any
detectable modifications of minichromosomes.
Structural Characterization of Artificial
Minichromosomes
The structure of the HACs was investigated by re-
striction digestion of genomic DNA from AG1-1, AG3-
1, AG6-1, and AP1-1, followed by PFGE and hybridi-
Mejı́a et al.: HAC-Mediated Rescue of HPRT Deficiency 321

zation with the 17 a–cDNA and HPRT1-cDNA probes
(fig. 4). Three enzymes with a known restriction pattern
on JM2860—NotI, SalI, and BssHII—as well as a com-
bination of BanII and EcoO109I were used.
JM2860 is cleaved, by NotI, at two closely spaced
sites flanking the boundary between the 17 a and HPRT1
genomic inserts (fig. 1A), generating two fragments—
one of 404 kb (detected by both probes; fig. 4A) and
one of !100 bp, which was not seen on the PFGE. Com-
pared to the parental HT1080 control cells, the 17 a
and the HPRT1 probes detected, in the four minichro-
mosome cell lines, additional DNA fragments from 50
kb to 11 Mb in size, indicating rearrangement of the
input DNA. Variation in the intensity of the bands
within a cell line probably reflects differences either in
copy number or in the relative contents of the target
sequences detected by either probe. Each cell line ex-
hibited a different NotI pattern.
Irregular rearrangement of input DNA from JM2860
in the minichromosome cell lines was also indicated by
analysis with BssHII, SalI (fig. 1A), and BanII⫹EcoO109I,
which do not cleave the 17 a DNA in JM2860. BssHII
and SalI generate 286-kb and 222-kb fragments, respec-
tively, in JM2860, as detected by the 17 a probe (fig. 4A),
whereas the HPRT1 gene lies within a 64-kb BssHII frag-
ment and a 114-kb SalI fragment, as detected by the
HPRT1 probe (fig. 4A). In most of the minichromosome
cell lines, BssHII and SalI fragments similar in size to those
from JM2860 are observed, but these are part of a com-
plex pattern including smaller and larger fragments de-
tected with either probe and absent from the parental
HT1080 cell line (fig. 4A). Although larger fragments may
result from inactivation of relevant restriction sites by
overlapping CpG methylation of genomic DNA, these
results are compatible with the irregular, alternating pat-
tern of alphoid and HPRT1 segments observed by fiber
FISH.
Used in combination, BanII and EcoO109I generate a
220-kb JM2860 fragment detected by the 17 a probe (fig.
4B), whereas the nonalphoid moiety of the construct is
digested to fragments !1 kb on average. Compared to the
parental HT1080 control, the size of BanII⫹EcoO109I
fragments from the minichromosome cell lines detected
by the 17 a probe was ∼50–250 kb (fig. 4B). Both en-
donucleases are insensitive to CpG methylation. Under
the assumptions that each of the fragments detected oc-
curs once on the minichromosome and that together they
represent an alphoid content similar to that in the parental
JM2860 (55%), the order of magnitude for minichromo-
some size is estimated at 1–5 Mb.
To determine whether the HACs were linear or cir-
cular structures, DNA from AG1-1, AG3-1, AG6-1,
and AP1-1 was digested by the exonuclease Bal31, fol-
lowed by XhoI restriction cleavage. Filter transfers

Figure 4 Structural analysis of minichromosome DNA. Genomic DNA from minichromosome clones AG1-1 (lanes 1), AG3-1 (lanes 2),
AG6-1 (lanes 3), and AP1-1 (lanes 4) and from HT1080 HPRT⫺ nontransfected cells (lanes C) was fractionated, after restriction-enzyme
digestion, by PFGE, as indicated. A, Cleavage by BssHII (lanes B), NotI (lanes N), and SalI (lanes S). A filter transfer was successively hybridized
with the 17 a BAC used in JM2860 and with an HPRT1-cDNA probe; the results are shown side by side. Similarities between the NotI patterns
suggest detection of the same fragments by either probe. The fourth column shows fragments cleaved by BssHII, NotI, and SalI of the parental
HAC construct JM2860 (fig. 1A). (The HPRT1-locus BssHII and SalI fragments of 64 and 114 kb, respectively, here display a lower apparent
size.) The positions of DNA size standards (l-DNA fragments and concatemers as well as Saccharomyces cerevisiae chromosomes) are indicated.
B, BanII⫹EcoO109I digestion. A filter transfer was hybridized with the 17 a probe, which detected a number of fragments of different sizes
in each cell line, indicating the length of the alphoid segments in the HACs. Several fragments of similar size were also shared by the four cell
lines and the parental HT1080 HPRT⫺ cells. The sizes (in kb) of the l-DNA concatemer standards are indicated.
322 Am. J. Hum. Genet. 69:315–326, 2001

were hybridized with probes for the BAC vector and Mitotic Stability of HACs
the ampicillin-resistance genes from either arm of
To investigate HAC stability in AG1-1, AG3-1, AG6-
JM2860 (fig. 1B), to detect truncated fragments re-
1, and AP1-1, these cell lines were analyzed for the per-
sulting from the removal, by the exonuclease, of se-
sistence of artificial minichromosomes during growth
quences from each telomeric end. Figure 5 shows that
under nonselective conditions. FISH was performed with
in each cell line fragments were generated that—
the 17 a probe after 26–40 d and again, after 54–62 d
compared to the I-SceI–linearized DNA from JM2860,
of culture in the absence of G418, puromycin, and HAT.
which was entirely degraded—were resistant to Bal31
The results of this analysis are summarized in table 1.
activity. When both the sizes of the fragments detected
Daily rates of minichromosome loss were calculated
and the restriction map of JM2860 are taken into ac-
from the number of minichromosome-positive meta-
count, these results show the presence of an internal
phase spreads; the values were .0049–.017 (0.49%–
JM2860 segment stretching from the BAC vector to the
1.7%/d), depending on the cell line, without large dif-
ampicillin-resistance gene. This may be due to a
ferences between the first and the second FISH analyses.
JM2860 molecule having escaped I-SceI cleavage. Al-
In two cell lines, AG1-1 and AG3-1, minichromosome
though linearization of the vector was confirmed by
loss was accompanied by integration events in the nat-
PFGE, residual circular molecules were not specifically
ural chromosomes, affecting 8%–16% of nuclei; the
detected, because these fail to fractionate into the gel.
proportion of these events, however, did not increase
An additional, rearranged internal copy of the same
during the second off-selection month. With a generation
segment was detected in AG6-1.
time of ∼1 d for HT1080 cells, these rates of minichro-
Further studies using controlled g irradiation and frac-
mosome loss suggest efficient segregation in 98%–99%
tionation, on a pulsed-field gel (Taylor et al. 1996), of
of mitoses and, therefore, the presence of a functional
of DNA molecules ⭐3 Mb in size suggested that the
centromere in all but a small minority of the minichro-
artificial minichromosomes were circular molecules
mosomes. The number of artificial minichromosomes
(data not shown). Both the 17 a probe and a specific
per nucleus was unchanged relative to the cells grown
probe containing the lacZ and ampicillin-resistance
in selective medium.
genes detected a band within the limiting mobility of the
pulsed-field gel—which was present only after g irra-
Complementation of the HPRT⫺ Phenotype
diation and increased in intensity with higher doses, in-
dicating that the circular molecules had generated linear The ability of HACs to serve as vectors for gene trans-
forms after breakage. fer was investigated by including the HPRT1 gene (40

Figure 5 Bal31 exonuclease analysis. Southern blot analysis of Bal31-digested genomic DNA shows that the telomeric ends of JM2860
(fig. 1A) are internal in the HACs. Zero (lanes 0), two (lanes 2), or four (lanes 4) units of Bal31 exonuclease were used, followed by digestion
by XhoI. The 4.8-kb and 6.3-kb terminal XhoI fragments of I-SceI–cleaved JM2860 are detected, respectively, by AmpR and repE probes (fig.
1B) in the control, in which the exonuclease is omitted (lanes 0). These fragments are degraded by Bal31, whereas the fragments detected by
the probes in minichromosome cell lines are resistant (notwithstanding a band-intensity decrease expected in light of the extension of Bal31
degradation to internal regions). The 6.8-kb JM2860 fragment detected by the AmpR probe probably results from incomplete digestion by I-
SceI. Genomic DNA from the parental HT1080 cells was included as a negative control for the minichromosome probes used.
Mejı́a et al.: HAC-Mediated Rescue of HPRT Deficiency 323

Table 1
Minichromosome Stability in the Absence of Selective Pressure
FISH ANALYSIS OF CELLS OFF SELECTIONb
(% of spreads)
MINICHROMOSOMES Other
OBSERVED UNDER Chromosomal DAILY RATE OF
CELL SELECTIONa DAYS OFF Terminal Integration No MINICHROMOSOME
LINE INITIAL SELECTION (% of spreads) SELECTION Minichromosome Integrationd Events Signale LOSSc
AG1-1 HAT 74 32 50 0 10 40 .012
62 32 0 8 60 .013
AG3-1 G418 ⫹ puromycin 88 40 44 0 16 40 .017
54 36 0 12 52 .016
AG6-1 Triple 96 32 82 0 0 18 .0049
58 68 0 0 32 .0059
AP1-1 Triple 92 32 60 0 0 40 .013
60 32 0 0 68 .017
a
Control cells had been passaged under selection for 7–11 wk since transfection.
b
FISH analysis of metaphase chromosome spreads was performed with a probe for 17 a DNA; 50 spreads from each cell line were analyzed
at the two time points indicated, after selective pressure was lifted.
c
Calculated by the formula Nn p N0 # (1 ⫺ R)n where N0 is the number of metaphase chromosome spreads showing minichromosomes in
the cells cultured under selection, Nn is the number of minichromosome-containing metaphase chromosome spreads after n days of culture in
the absence of selection, and R is the daily rate of loss.
d
Hybridization signal at the tip of a natural chromosome, suggesting a telomere-directed truncation event.
e
Spreads with a hybridization signal on the natural chromosomes 17 only.

kb) in JM2860 and by using host cells with an HPRT⫺
genetic background (Lugo and Baker 1983). Cell lines
containing minichromosomes were successfully isolated
in the presence of HAT supplement, to select for the
HPRT1 gene. The presence of HPRT1 FISH signals only
on the artificial minichromosomes and at the nonfunc-
tional locus in Xq26.2 demonstrated that one or more
copies of the HPRT1 gene linked to the minichromo-
somes in AG1-1, AG3-1, AG6-1, and AP1-1 must be
responsible for the complementation of the HPRT⫺ phe-
notype of the host cells. One of the cell lines, AG3-1,
was isolated and expanded in the absence of selective
pressure for the HPRT1 gene. Subsequent transfer to
growth medium containing HAT resulted in no adverse
effects for the culture, suggesting continued expression
of the gene.
Further evidence of the transformation to an HPRT⫹
phenotype mediated by the minichromosomes was ob-
tained by use of 6-TG, which is converted to cytotoxic
thioguanine deoxynucleotide triphosphate in an HPRT-
dependent pathway. Whereas the parental HT1080 cells
exhibited normal growth in the presence of 5 mg 6-TG/
ml, the four minichromosome lines showed complete cell
death in ⭐5 d. By contrast, cells of the four lines pre-
viously grown off selection for 57 d maintained high
densities and, over the same 5-d period, reached conflu-
ence in the presence of 6-TG. This is consistent with the
growth of a subpopulation of minichromosome-negative
cells during extended culture off selection (table 1). It
also indicates that minichromosome formation was not
accompanied by integrations of the HPRT1 gene into
the natural chromosomes, in accordance with the data
obtained by FISH analysis of the four cell lines.
Discussion
In this study, we intended to determine whether the in-
tegration of a full-length human genomic locus into a
HAC construct would allow the formation of artificial
chromosomes complementing the corresponding meta-
bolic deficiency in human cultured cells. A large HAC
construct containing human telomeres, 17 a DNA, and
the HPRT1 gene was transferred into HPRT⫺ HT1080
cells, and several aminopterin-resistant, positive clones
were isolated. These were shown to contain HACs with
a functional centromere at a consistently low copy num-
ber of 1 or 2/cell. The HACs were derived only from
the input DNA and contained one or possibly several
copies of the HPRT1 gene. Analysis of HAC stability
showed that accurate segregation occurred in 198% of
mitoses and that, in the absence of selective pressure,
35%–70% of the cells retained HACs at the end of a
2-mo period. These frequencies were comparable to rates
of minichromosome loss in HT1080 cells reported by
Ikeno et al. (1998). In two other studies (Harrington et
al. 1997; Ebersole et al. 2000), HAC transmission was
observed in 100% of mitoses of some clones, although
variation in mitotic stability was noted between different
cell lines.
This study has also shown that the HACs obtained
were several megabases in size, which is significantly
larger than the input DNA derived from the HAC con-
324
struct, JM2860. The studies on DNA structure and fi-
ber-FISH analysis indicated that the HACs consist of
irregular alternating stretches of alphoid and HPRT1-
locus DNA. This may have resulted from recombination
of the input DNA, ligation of random truncated frag-
ments, or a combination of these mechanisms. Southern
hybridization data from our Bal31 analysis indicate that
certain minichromosome segments probably derive
from JM2860 molecules that have escaped digestion by
I-SceI endonuclease prior to transfection. Except for the
relatively small restriction fragments detected in that
experiment, the linear or circular origin of minichromo-
some DNA cannot be confirmed. Chen et al. (2001)
found that linear but not circular plasmid molecules
formed large concatemers, possibly by nonhomologous
end-joining (Critchlow and Jackson 1998), when trans-
ferred to mouse liver cells in vivo. Head-to-tail conca-
temers integrated into the host-cell chromosomes, how-
ever, have been observed with both linear and circular
DNA (Folger et al. 1982). The irregular structure of the
minichromosomes in the present study may be related
to other structural aspects of input DNA, such as the
inclusion of a large noncentromeric insert in JM2860.
Under the assumption of an unbiased representation
of the alphoid and nonalphoid components of the HAC
vector in the minichromosomes, these may be expected
to contain several copies of the HPRT1 gene, although
the number of intact, potentially functional copies is
difficult to determine. Given the nature of the final HAC
structure, our results indicate that, in a context in which
both transgene copy number and expression levels must
be tightly controlled, the insertion, by recombinational
methods, of the transgene into an established HAC (Ku-
roiwa et al. 2000) should be considered.
HACs have a distinct advantage over those gene-
transfer vectors that either may be silenced following
integration into the host-cell chromosomes or may in-
corporate cDNA inserts that may lack important se-
quences for long-term regulated expression of the trans-
gene. By relying entirely on the endogenous mechanisms
for chromosome maintenance, HACs also offer a valid
alternative approach to viral vectors such as adenovirus.
Recently, Wade-Martins et al. (2000), using an episomal
vector derived from Epstein-Barr virus (EBV), reported
the transfer of the HPRT1 locus. Although this is a
useful strategy for both the expression and the analysis
of large genes in some mammalian cell lines, the system
may not be generally applicable, because EBV-based
vectors rely on the presence of a viral transactivator,
EBNA1, and because there are safety concerns over their
use for gene therapy (Van Craenenbroeck et al. 2000).
Moreover, in the aforementioned report, the rates of
episome loss in the absence of selection were fourfold
greater than those for HACs in the present study.
Previous studies in this field have aimed to maximize
Am. J. Hum. Genet. 69:315–326, 2001
the size of the alphoid-DNA fragments used to generate
artificial chromosomes, either by preparing concatemers
of an alphoid insert (Harrington et al. 1997) or by work-
ing with a large alphoid yeast-artificial-chromosome
clone (Henning et al. 1999). It is unclear, however,
whether a large continuous array of alphoid DNA (a)
is necessary to form a functional minichromosome cen-
tromere or (b) increases the probability that the centro-
mere will initially be established. In the present study,
the irregular arrangement of alternating alphoid and
nonalphoid sequences observed in the HACs conflicts
with the structure of natural centromeres, in which con-
tinuous alphoid-DNA arrays may be up to several me-
gabases in size. The largest 17 a–DNA fragments in our
HACs were 158–250 kb in size, providing an estimate
of the minimal length that may be required to form a
de novo centromere (however, two closely spaced al-
phoid segments could possibly act as a single functional
unit [Choo 2000]).
Recent preliminary data from our laboratory, how-
ever, indicate both that mitotic stability is not strictly
correlated with the size of continuous alphoid segments
and that the exclusion of the HPRT1 insert from the
parental HAC construct results in lower minichromo-
some loss (J. E. Mejı́a, A. Alazami, A. Willmott, and
Z. Larin, unpublished data). It is possible that, in the
JM2860 HACs, the boundaries between alphoid and
nonalphoid DNA fail to determine the boundaries be-
tween heterochromatic and euchromatic DNA. Early S-
phase origins of replication in transcriptionally active
segments derived from the HPRT1 locus may initiate
premature replication of neighboring alphoid DNA.
Temporally decoupling the synthesis of alphoid DNA
from that of centromeric proteins such as centromeric
protein A (Karpen and Allshire 1997; Choo 2000) could
impair the maintenance of functional centromeres in
HACs. Conversely, longer stretches of alphoid DNA
may preserve internal segments of late replication, by
virtue of a greater distance between these and early S-
phase origins of replication. Replication interference
may inhibit kinetochore formation in a portion of ar-
tificial minichromosomes, leading to nonattachment to
the mitotic spindle and, at anaphase, the failure to seg-
regate with the other chromosomes. This mechanism of
minichromosome loss is consistent with the conserva-
tion of minichromosome-copy number seen in the re-
maining positive cells. It may not be a coincidence that
the lowest rate of minichromosome loss observed in the
present study occurs in the cell line (AG6-1) in which
the maximal-alphoid-segment size is the longest.
Future studies could address the timing of replication
of both alphoid and nonalphoid DNA in artificial chro-
mosomes—and its relation to the maintenance and sta-
bility of artificial chromosomes in human cells. Should
the detrimental effect that replication-timing interfer-
Mejı́a et al.: HAC-Mediated Rescue of HPRT Deficiency
ence may have on centromere function be demonstrated,
it might be possible to isolate the alphoid-DNA segment
in the HAC construct by flanking it with replication-
timing–boundary elements (Watanabe et al. 2000). The
use of a shorter alphoid segment in an optimal sequence
context would allow a reduction in the total size of HAC
constructs, to make them compatible with more-effi-
cient delivery vehicles, such as the HSV-1 amplicon
packaging system, with a theoretical transgene capacity
of ⭐150 kb (Sena-Esteves et al. 2000).
We have shown that artificial chromosomes based on
the major component of natural human centromeres,
alphoid DNA, can be used as vectors for the comple-
mentation of a genetic deficiency in human cells. The
HACs generated in this study constitute a pertinent
model to address the formation and maintenance of de
novo centromeres and indicate directions for further
research and improvements in HAC-vector technology
for gene transfer.
Acknowledgments
This work was supported by a fellowship from the Well-
come Trust (to Z.L.), by SmithKline Beecham, by the Uni-
versity of Oxford Medical Research Fund, and by the Dys-
trophic Epidermolysis Bullosa Research Association of the
United Kingdom.
Electronic-Database Information
Accession numbers and the URL for data in this article are
as follows:
Online Mendelian Inheritance in Man (OMIM), http://www
.ncbi.nlm.nih.gov/Omim/ (for HPRT1 [MIM 308000] and
Lesch-Nyhan syndrome [MIM 300322])
Project Ensembl, http://www.ensembl.org/ (for HPRT1 [En-
sembl gene ENSG00000101965])
References
Boussif O, Lezoualc’h F, Zanta MA, Mergny MD, Scherman
D, Demeneix B, Behr J-P (1995) A versatile vector for gene
and oligonucleotide transfer into cells in culture and in vivo:
polyethylenimine. Proc Natl Acad Sci USA 92:7297–7301
Caskey CT, Kruh GD (1979) The HPRT locus. Cell 16:1–9
Chen Z-Y, Yant SR, He C-Y, Meuse L, Shen S, Kay MA (2001)
Linear DNAs concatemerize in vivo and result in sustained
transgene expression in mouse liver. Mol Ther 3:403–410
Choo KHA (2000) Centromerization. Trends Cell Biol 10:
182–188
Critchlow SE, Jackson SP (1998) DNA end-joining: from yeast
to man. Trends Biochem Sci 23:394–398
Ebersole TA, Ross A, Clark E, McGill N, Schindelhauer D,
Cooke H, Grimes B (2000) Mammalian artificial chromo-
some formation from circular alphoid input DNA does not
require telomere repeats. Hum Mol Genet 9:1623–1631
Folger KR, Wong EA, Wahl G, Capecchi MR (1982) Patterns
325
of integration of DNA microinjected into cultured mam-
malian cells: evidence for homologous recombination be-
tween injected plasmid DNA molecules. Mol Cell Biol 2:
1372–1387
Harrington JJ, Van Bokkelen G, Mays RW, Gustashaw K, Wil-
lard HF (1997) Formation of de novo centromeres and con-
struction of first-generation human artificial microchromo-
somes. Nat Genet 15:345–355
Henning KA, Novotny EA, Compton ST, Guan X-Y, Liu PP,
Ashlock MA (1999) Human artificial chromosomes gener-
ated by modification of a yeast artificial chromosome con-
taining both human alpha satellite and single-copy DNA
sequences. Proc Natl Acad Sci USA 96:592–597
Ikeno M, Grimes B, Okazaki T, Nakano M, Saitoh K, Hoshino
H, McGill NI, Cooke H, Masumoto H (1998) Construction
of YAC-based mammalian artificial chromosomes. Nat Bio-
technol 16:431–439
Karpen GH, Allshire RC (1997) The case for epigenetic effects
on centromere identity and function. Trends Genet 13:489–
496
Kim U-J, Birren BW, Slepak T, Mancino V, Boysen C, Kang
H-L, Simon MI, Shizuya H (1996) Construction and char-
acterization of a human bacterial artificial chromosome li-
brary. Genomics 34:213–218
Kuroiwa Y, Tomizuka K, Shinohara T, Kazuki Y, Yoshida H,
Ohguma A, Yamamoto T, Tanaka S, Oshimura M, Ishida I
(2000) Manipulation of human minichromosomes to carry
greater than megabase-sized chromosome inserts. Nat Bio-
technol 18:1086–1090
Larin Z (1995) Functional analysis of mammalian genomes
using yeast artificial chromosomes. In: Monaco AP (ed)
Pulsed field gel electrophoresis—a practical approach. IRL
Press, Oxford, pp 139–157
Lugo TG, Baker RM (1983) Chromosome-mediated gene
transfer of HPRT and APRT in an intraspecific human cell
system. Somat Cell Genet 9:175–188
Marschall P, Malik N, Larin Z (1999) Transfer of YACs up
to 2.3 Mb intact into human cells with polyethylenimine.
Gene Ther 6:1634–1637
Mejı́a JE, Larin Z (2000) The assembly of large BACs by in
vivo recombination. Genomics 70:165–170
Mejı́a JE, Monaco AP (1997) Retrofitting vectors for Escherichia
coli-based artificial chromosomes (PACs and BACs) with
markers for transfection studies. Genome Res 7:179–186
Mills W, Critcher R, Lee C, Farr CJ (1999) Generation of an
approximately 2.4 Mb human X centromere-based mini-
chromosome by targeted telomere-associated chromosome
fragmentation in DT 40. Hum Mol Genet 8:751–761
Rasheed S, Nelson-Rees WA, Toth EM, Arnstein P, Gardner
MB (1974) Characterization of a newly derived human sar-
coma cell line (HT-1080). Cancer 33:1027–1033
Saitoh H, Tomkiel J, Cooke CA, Ratrie H, Maurer M, Roth-
field NF, Earnshaw WC (1992) CENP-C, an autoantigen in
scleroderma, is a component of the human inner kinetochore
plate. Cell 70:115–125
Sena-Esteves M, Saeki Y, Fraefel C, Breakefield XO (2000)
HSV-1 amplicon vectors—simplicity and versatility. Mol
Ther 2:9–15
Sullivan BA, Schwartz S (1995) Identification of centromeric
antigens in dicentric Robertsonian translocations: CENP-C
326
and CENP-E are necessary components of functional cen-
tromeres. Hum Mol Genet 4:2189–2197
Taylor SS, Larin Z, Tyler-Smith C (1994) Addition of functional
human telomeres to YACs. Hum Mol Genet 3:1383–1386
——— (1996) Analysis of extrachromosomal structures con-
taining human centromeric alphoid satellite DNA sequences
in mouse cells. Chromosoma 105:70–81
Van Craenenbroeck K, Vanhoenacker P, Haegeman G (2000)
Episomal vectors for gene expression in mammalian cells.
Eur J Biochem 267:5665–5678
Wade-Martins R, White RE, Kimura H, Cook PR, James MR
(2000) Stable correction of a genetic deficiency in human
cells by an episome carrying a 115 kb genomic transgene.
Am. J. Hum. Genet. 69:315–326, 2001
Nat Biotechnol 18:1311–1314
Watanabe Y, Tenzen T, Nagasaka Y, Inoko H, Ikemura T
(2000) Replication timing of the human X-inactivation cen-
ter (XIC) region: correlation with chromosome bands. Gene
252:163–172
Willard HF (1998) Centromeres: the missing link in the de-
velopment of human artificial chromosomes. Curr Opin Ge-
net Dev 8:219–225
——— (2000) Artificial chromosomes coming to life. Science
290:1308–1309
Yang JW, Pendon C, Yang J, Haywood N, Chand A, Brown
WRA (2000) Human mini-chromosomes with minimal cen-
tromeres. Hum Mol Genet 9:1891–1902